Google will pay SpaceX $920M per month for compute
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Google has agreed to pay SpaceX $920 million monthly for AI compute resources from October 2026 to June 2029, involving about 110,000 NVIDIA GPUs and other hardware. This deal highlights Google’s expanding AI infrastructure and SpaceX’s growing role in AI data centers ahead of its IPO.

SpaceX has confirmed a deal with Google in which the tech giant will pay $920 million per month starting in October 2026 for access to approximately 110,000 NVIDIA GPUs, CPUs, and related hardware. The agreement, announced via a regulatory filing, underscores Google’s surge in AI demand and SpaceX’s expanding role in AI infrastructure as it prepares for its historic IPO.

The deal spans from October 2026 through June 2029, with Google paying SpaceX $920 million each month for compute resources. SpaceX did not specify which data center Google will use, but the company’s CEO has previously indicated that the Colossus 2 data center near Memphis may be reserved for its AI unit, xAI.

According to the filing, the hardware involved includes roughly 110,000 NVIDIA GPUs, CPUs, memory, and other components. The agreement features a cancellation clause allowing either party to terminate with 90 days’ notice after December 31, 2026, with provisions for ramp-up and fee adjustments if access targets are not met by September 2026.

Why It Matters

This deal signifies a major expansion of Google’s cloud AI infrastructure, reflecting its need to meet surging customer demand for products like Gemini Enterprise. It also highlights SpaceX’s emerging role as a key provider of AI compute capacity, which could influence the broader cloud and AI hardware markets. The deal’s scale and timing suggest both companies are positioning for significant growth ahead of SpaceX’s IPO, which aims to raise approximately $75 billion at a valuation of around $1.75 trillion.

Amazon

NVIDIA GPU server hardware

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Background

In May, SpaceX announced a similar deal with AI startup Anthropic, paying $1.25 billion monthly for access to its Colossus 1 data center near Memphis. That deal was part of SpaceX’s strategy to monetize its AI infrastructure ahead of its IPO. Google’s recent demand for AI compute is driven by its expanding AI product portfolio, including the Gemini platform, which has seen higher-than-expected adoption.

Google is already considered one of the largest owners of AI compute globally, and its investment in this deal indicates a continued push to scale its cloud infrastructure. The company’s parent, Alphabet, has committed over $180 billion in capital expenditures this year, with plans to increase spending in 2027, including an $80 billion equity sale announced recently.

“This is a short-term, timely agreement to ensure we have bridge capacity to meet surging customer demand for our agent platform, Gemini Enterprise.”

— a Google representative

“The scale of this deal shows how big AI compute needs are becoming, and SpaceX’s emerging role as a provider in this space.”

— an anonymous researcher

Amazon

AI compute hardware for data centers

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

What Remains Unclear

It is not yet clear which specific data center Google will use or how the deal’s ramp-up will proceed. The exact impact on SpaceX’s overall data center strategy and the future of its orbital data center plans remain uncertain. Details about potential long-term extensions or modifications to the agreement are also still developing.

Amazon

high performance CPUs for AI

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

What’s Next

Next steps include Google ramping up its access to the hardware through September 2026, with potential adjustments based on initial deployment. The companies may also negotiate further terms or expand cooperation, especially around the development of orbital data centers. SpaceX’s IPO, scheduled for later in 2026, will be a key milestone to watch, as it could influence the company’s capacity to expand its AI infrastructure offerings further.

Amazon

enterprise GPU storage solutions

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

Why is Google investing so heavily in AI compute capacity?

Google’s investment reflects the rapid growth in demand for AI products and services, including its Gemini platform, which requires substantial compute resources to operate at scale.

What does this deal mean for SpaceX’s future plans?

The deal indicates SpaceX’s emerging role as a major provider of AI infrastructure, positioning it to capitalize on the expanding cloud AI market ahead of its IPO.

Will this impact other cloud providers or AI hardware suppliers?

Potentially, as it signals a significant shift towards large-scale AI compute deals, which could influence market dynamics among cloud providers and hardware manufacturers.

Is the deal exclusive or could SpaceX partner with other companies?

The agreement includes a cancellation clause allowing either party to terminate with 90 days’ notice, so the arrangement may not be exclusive long-term, and SpaceX could pursue additional partnerships.

Source: Hacker News

You May Also Like

Sony’s new Xperia phone gets an overdue redesign

Sony unveils the Xperia 1 VIII with a new square camera island, larger telephoto sensor, and updated features, marking a significant aesthetic update.

China’s Loongson launches homegrown 16-core server CPU built on LoongArch architecture — 40W chip with DDR4 ECC and 32 PCIe lanes targets cheap SMB file, database, and web servers

Loongson announces the Loongson 3C3000, a 16-core server processor based on LoongArch architecture, targeting low-cost business workloads.

Cordless Vacuum Basics: Power, Airflow, and What to Ignore

Understanding cordless vacuum basics—power and airflow—reveals what truly impacts cleaning performance, so keep reading to learn what features to ignore.

How Fast Does Claude, Acting as a User Space IP Stack, Respond to Pings?

A test measures how quickly Claude, acting as a user space IP stack, responds to ICMP ping requests, revealing insights into AI-driven network processing.